ECON3173 – Cross Section and Panel Data Analysis

Individual Project: Guidelines and Questions

This document provides guidelines and questions for the Individual Project of ECON3173, which accounts for 40% of the total marks.

Honoring the precepts of academic integrity and applying its principles are fundamental responsibilities of all students and scholars at UIC. You are advised to read through the ‘UIC Guidelines for Handling Academic Dishonesty’ file on iSpace before you start your assignment. Any form. of plagiarism or cheating can result in various disciplinary and corrective activities. Using generative AI tools is not allowed.

Deadline: by 20/12/2024. Submission Method:

a)    Please   submit  your  typing  assignment  report  in  a  single   PDF  file  to  Turnitin ‘Submission Link: Report’via iSpace. The filename of your PDF submissions should have  the  following  format: ECON3173_Project_Student  ID_Name  in  Pinyin   (e.g., ECON3173_Project_190000001_Mi Lin).

b)    Save    your    data     and    .do    file(s)     in    a     zip    file.    Name     your    zip    file     as ECON3173_Project_Student ID_Name in Pinyin. Then, upload your file to‘Submission Link:  Stata  Data  and  Program’via  iSpace.  You  are  expected  to  submit  2  .do  files,

namely ParA.do and PartB.do, respectively, should be able to replicate each part of your submitted work.

c)    Use the ‘ECON3173_Individual  Project_Report Template’ file on the iSpace to input your report. Ensure you provide a question number for each part of your work.

Format Requirements

Part A: Imitate an existing research (30%)

Traffic crashes are the leading cause of death for Americans between the ages of 5 and 32. Through various  spending  policies,  the  federal  government  has  encouraged  states  to institute mandatory seat belt laws to reduce the number of fatalities and serious injuries. In this exercise, following Einav and Cohen (2003), you will investigate how effectively these laws increase seat belt use and reduce fatalities using the “SeatBelts.dta” dataset posted on iSpace. The data file Seatbelts contains a panel of data from 50 U.S. states plus the   District   of   Columbia    from   1983   through    1997.    The   dataset   is    detailed   in “Seatbelts_Description.pdf”.  It  was  used  in the  Einav  and  Cohen  (2003)  paper,  which serves as the background reading for this exercise.   Both files are available on iSpace as well.

A1.    Using   OLS  to  estimate  the  effect  of  seat   belt  use  on  fatalities  by  regressing fatalityrate on sb_useage, speed65, speed70, ba08, drinkage21, ln(income), and age. Does the estimated result suggest that increased seat belt use reduces fatalities? Report, interpret, and comment on your results. (5%)

A2.    Run  a  one-way  fixed  effect model with state-fixed  effects.  Do the results  change when you add state-fixed effects? Run a two-way fixed effects model with both time and state-fixed effects. Do the results change when you add time-fixed effects plus state-fixed effects? Report, interpret, and comment on your results. (5%)

A3.    Which  regression  specification  you  obtained  from  A1  and  A2  is  most  reliable? Explain why. (5%)

A4.    Using the results from the two-way fixed effects model with both time and state- fixed effects, discuss the magnitude of the coefficient on sb_useage. Is it large? Small? How  many  lives  would  be  saved  if  seat  belt  use  increased  from  52%  to  90%? Illustrate your calculation and comment on your result. (5%)

A5.    There  are  two  ways  that  mandatory  seat  belt  laws  are   enforced:  “Primary” enforcement means that a police officer can stop a car and ticket the driver if the  officer  observes  an  occupant  not  wearing  a  seat  belt;  “secondary”  enforcement  means that a police officer can write a ticket if an occupant is not wearing a seat belt,  but must have another reason to stop the car. In the data set, primary is a binary  variable for primary enforcement, and secondary is a binary variable for secondary  enforcement. Run a regression of sb_useage on primary, secondary, speed65, speed70,  ba08, drinkage21, ln(income), and age, including fixed state and time effects in the  regression.  Does  primary  enforcement  lead  to  more  seat  belt  use?  What  about  secondary enforcement? Report, interpret, and comment on your results. (5%)

A6.    In 2000, New Jersey changed from secondary enforcement to primary enforcement. Assume that data availability is not an issue, design a Differences-in-Differences estimation strategy to estimate the number of lives potentially saved per year by making this change. Explain your approach. (5%)

Part B: A small-scale research project – innovation (70%) Introduction:

In this project, you are invited to empirically investigate potential causality between firms’ business  environment  and  economic  performance  using  a  firm-level  dataset  from  economies  included  in  the  World  Bank  Enterprise  Survey  Data  (WBESD).  WBESD  database collects information about an economy’s business environment, how individual  firms  experience  it,  how  it  changes  over  time,  and  the  various  constraints  to  firm  performance and growth, etc. The entire database is available to researchers and includes  all questions from the surveys at the firm level.

Guidelines to download and prepare data for this individual project:

a)   Please visit https://login.enterprisesurveys.org/ to register your user account for the WBESD database (see the snapshot below). Registration is free.

b)  There are a total of 97 economies represented in the World Bank Enterprise Surveys Database (WBESD). Among these, 61 economies have a time span of at least three years. For their individual projects, students are required to select data from a panel of random combinations of three different economies out of these 61 economies.

Data allocation protocol: Students are required to pick a lottery ticker number first. An “Individual Project Lottery Ticket Sign-up Sheet” will be available in iSpace from 9 p.m. on Tuesday, 26/11/2024. Please sign up for a lottery ticket number by 12 noon on Wednesday, 27/11/2024. We will operate on a 'first-come, first-served' basis.

A lucky draw will be conducted in class on Thursday, 28/11/2024, to assign specific economies to each lottery ticket number.

c)   Once registration is completed, login and download the data following the steps below:

i.     Login with yourusername and password. You will be directed to the ‘Full Survey Data’ page.

ii.     Select ‘Panel data’ under ‘Survey Type’ on the left. Ensure you are on the ‘Data by Economy’ view instead of ‘Combined Data’ . See the snapshot below.

iii.     Download  your  economies’  corresponding  data and documentation for all the available years.

For example, Afghanistan has two panel data files, one for 2005 and 2009 and the other for 2008, 2010, and 2014. Then download both of them.

iv.     Extract the data and survey documentation files into a working folder on your PC. Now the data file is ready to open in Stata.

Answer ALL of the Following Questions

Note that this is not an essay-type assignment. Please answer the questions one by one. For each question, the performance of the Stata do files accounts for 20% of the marks.

B1)   Use the Stata command “append” to append data of all years and economies into a single Stata data file with panel data format. Select and rename variables in Table 1. ‘Old name’ refers to the variable name in the original dataset, while ‘New name’ is the new  corresponding name to be  defined.  Generate  a  new  variable  exp_dum (export  dummy)  equals  1  if sales_exp  is  positive;  otherwise,  0.  Generate  a  new variable foreign_dum  (foreign  ownership  dummy)  equals  1 if foreign is positive; otherwise, 0. Generate a new variable soe_dum (state ownership dummy) equals 1 if soe is positive; otherwise, 0. (5%)

Table 1: Variable List

B2)   Conduct exploratory data analysis for variables listed in B1), i.e., using appropriate summary statistics (e.g., observation number, mean, standard deviation, minimum and maximum values, etc.) to explain your data and make necessary comments.

(10%)

Considering total  output is  measured by Sales, labour input is measured by the total number  of  employees,   capital  is  measured  by  capital,  and  intermediate  inputs  are measured by materials, a production function can be written as:

sales = Acapitalα Employeesβ Materialsy                                                             (1)

where A is the total factor productivity (TFP). Based on panel data, taking logarithms on both sides, equation (1) is transformed to

ln(salesit ) = ln(A) + α ln(capitalit ) + β ln(Labourit ) + y ln(Materialsit ) + eit           (2)

B3)   Based  on the variable ‘panelid’,  set the dataset in a panel format, then obtain the estimated coefficients in equation (2) by running panel data regression. Comment on  your  results,  including  a)  a  discussion  on  the  capital,  labour,  and  materials elasticities of sales, respectively, and b) a comparison between the panel two-way fixed effect and random effect model. (10%)

B4)   Include  Export_dumit    as a new variable of interest in equation (2) to test if a firm’s   performance improves after entering export markets. Based on relevant economic   theories or literature, explore the data set to add appropriate control variables to   the production equation (2). Run a panel regression and interpret the results. (10%)

B5)   To what extent could we use the estimated coefficient on  Export_dumit    obtained in B4) for causal inference? Explain. Illustrate an appropriate empirical strategy to make improvements if deemed required. Finally, reestimate the model based on your proposed empirical strategy, compare the results to what you have obtained in B4, and comment on the results. (10%)

B6)   Explore the complete data set (i.e., do not have to be limited to the variables listed  in B1) and design an empirical model to evaluate the impact of ownership structure  on the export decision. Interpret and comment on the empirical results you obtained.

(10%)

B7)   Predict  firm-level  productivity  (i.e.,  ln(A) + eit ) to test the  hypothesis that good business-government relations boost productivity. Explore the complete data set (i.e., do not have to be limited to the variables listed in B1) to propose an empirical model with an appropriate empirical strategy based on relevant economic theories or  literature.  Explain  the  variable(s)  you  select  for  this  question.  Interpret  and comment on the empirical results you obtained. (15%)